report generation problems - plsgui
gilles.plourde@mcgill.ca
Posted on 01/25/08 12:55:34
Number of posts: 23
Gilles
Many attempts were needed to get cluster reports for PET data. I am still trying to get a report for the second LV. I get the message "Generating cluster report - please wait" and
nothing may happen for hours after which there is freeze or a disconnect from the server. The process seems to work a bit better for the bootstrap ratios than for the saliences ("LV view").
The attempts were done on the Unix Brain Imaging Center network running Matlab 6. The data files includes 4 subjects and 4 periods. Min cluster size set to 100.
The files will be on http://www.bic.mni.mcgill.ca/~anast/ in a few minutes should there be a need to look at them.
Thanks!
Gilles
Untitled Post
I'm Online
jshen
Posted on 01/25/08 17:39:32
Number of posts: 291
It's expected that cluster report for bootstrap ratios will be faster than for that for saliences. This is because usually BSR has smaller cluster area.
For the same reason, in order to display cluster report for saliances, you would have to change the threshold to reduce the cluster area.
Untitled Post
I'm Online
nlobaugh
Posted on 01/28/08 09:48:22
Number of posts: 229
A better option for generating an image of clustered saliences is to do the following:
1) Generate a cluster report based on the bootstrap ratios, and save the results as a *_*cluster.mat file.
2) switch to BrainLV view
3) load the saved cluster report and use this to mask the saliences.
With this approach, the focus remains on stable saliences, not just strong saliences.
This issue is discussed in more detail in McIntosh&Lobaugh, 2004, Neuroimage,23:S250